Prateek057 / ADDClusterer

Architectural Design Decisions Clusterer

Workbench for Learning Similar Architectural Design Decisions(Service only)

Service for learning and retrieving similar Architectural Design Decisions

Runtime dependencies:

JDK - 1.8.0
MongoDB - 3.2.10

Core Plugin dependencies (check build.sbt)

Play - 2.5.9
Spark (core, sql, mllib) - 2.0.1
Morphia - 1.2.1
Angular - 1.5.8
Bootstrap - 3.3.6

Experimental

1. DL4J Version: 0.9.1
2. RapidMiner: Snapshot version to be build locally

Deploying the project

Ensure mongodb is running on default port 27017; else configure settings in Global.java
From the command prompt execute

sbt run

Running Applications

1. Frontend Application will be available at localhost:9000
2. Spark Web UI is available at localhost:9090

Example Training Pipelines

Spark KMeans

Word2Vec: localhost:9000/spark/kmeans/example1
HashingTF: localhost:9000/spark/kmeans/example2

DL4J KMeans

Word2Vec: localhost:9000/dl4j/kmeans/example1

Results

Results are stored in "myresources" directory. two type of results are generated currently

1. Labeled Data in json format, Stored in /myresources/results/
2. Fitted Model from the pipeline, Stored in /myresources/models/